This Page Is Inserted by IFW Operations
and is not a part of the Official Record
BEST AVAILABLE IMAGES
Defective images within this document are accurate representations of the
original documents submitted by the applicant.
Defects in the images may include (but are not limited to):
• BLACK BORDERS
• TEXT CUT OFF AT TOP, BOTTOM OR SIDES
• FADED TEXT
• ILLEGIBLE TEXT
• SKEWED/SLANTED IMAGES
• COLORED PHOTOS
• BLACK OR VERY BLACK AND WHITE DARK PHOTOS
• GRAY SCALE DOCUMENTS
IMAGES ARE BEST AVAILABLE COPY.
As rescanning documents will not correct images,
please do not report the images to the
Image Problems Mailbox.
WIS PAGE BLANK
WORLD INTELLECTUAL PROPERTY ORGANIZATION
International Bureau
PCT
INTERNATIONAL APPLICATION PUBLISHED UNDER THE PATENT COOPERATION TREATY (PCT)
(§11) Unteraat kraal Patent Classification 4 :
©Q)6F9>/38 9 US/06
At
{U) International PoabDicatSon Number: WO 86/ 07174
(43) nnternatt nal PuMicati n Bate: 4 December 1986 (04.12.86)
(211) IlQiteraatiointalt Application Number: PCT/US85/01435
(22) natteraatfioiaai Hlimg Pate : 29 July 1 985 (29.07.85)
<3E)> Priority Application Nmmnber: 735,641
(32) Priority ©ate: 20 May 1985 (20.05.85)
(33) Priority Country: US
(7fl)f72) Applicant omaS Inventor: SHEKELS, Howard, D.
[US/US]; 8619 N. Cardinal Drive, Phoenix, AZ 85028
(US).
(74) Agent: PHILLIPS, James, H.; Cates & Phillips, 3800 N.
Central Avenue, Suite 920, Phoenix, AZ 85012 (US).
(81) Designated States: DE (European patent), FR (Euro-
pean patent), GB (European patent), J P.
Published!
With international search report.
Before the expiration of the time limit for amending the
claims and go be republished in the event of the receipt
of amendments.
MAT.
> BOSSIER
(54) Title: SUPER-COMPUTER SYSTEM ARCHITECTURES
(57) Abstract
A computer system in which instruction sequencing is
under the control of a program control computer (2), but each
individual instruction is assigned for execution to a individual
instruction computer (5-10) in a bank of instruction compu-
ters. Each instruction computer includes programmable in-
struction decoding means (16) by the modification of which
the microtasks undertaken to execute an instruction are accor-
dingly modifiable, either between successive execution cycles
or during a single execution cycle. Thus, the system has the in-
herent ability to learn and adapt during the performance of a
pr gram. The system may be extended in two-dimensional
and three-dimensional arrays to obtain a multiplication of
power and to enjoy redundancy advantages.
Ststcu
BAW
1^
IfCUT
CQMSOLC
pqqcham cowtbol computer
I W STRUCT tO H ft OPERAND RECI5TCB IS) U
INSTRUCTION exCCUTttWg
COORDINATING UttTT H
ARITHMETIC &
LOGIC UNIT
IBSTftUCTOM
S££?UTCP OO
I UitTRUCTtON i f
CUS 3TSTCH_
. _ ibstruchonI i instruction I
COCgtfTCRff t ICOOPUTCR Qg| I COMPUTER 03 I * * *
7
ItBTCUCTtOW CSSt&aXttT fittD QJgRANO C3615TER tSI
P(W3 STSTgJ
n»a. eoctsoue-
I
taSTRuCTtON SECOOCNC
W5TQUCTION
tJJCROTASK
QCMCRATOA
OCHTKMCTtC AND
L0OC UMT
STATUS
JSDOC1D: <WO 86071 74A1_I_>
POR MUE FmPg»SB$ ®F ENfflSMMMON ONLY
Codes oased to ndcfiaiitfy States pasty to fclhe PCTon the front pages of pamphlets publishing international appli-
cations under the POT.
ait
Austria
<EA
Gabon
MR
Mauritania
Australia
<GEJ
United Kingdom
MW
Malawi
m
Barbados
Mil
Hungary
NL
Netherlands
Belgium
DTT
Htaly
wo
Norway
0G
Bulgaria
JUP
Japan
ao
Romania
OS
Brazil
KIP
Democratic People's Republic
SB
Sudan
ra-
Central African Republic
of Korea
SE
Sweden
ce
Congo
m
Republic of Korea
SN
Senegal
CSS
Switzerland
U
Liechtenstein
su
Soviet Union
CM
Cameroon
IUt
Sri Lanka
* TD
Chad
©E
Germany, Federal Republic of
Luxembourg
TO
Togo
OS
Denmark
MC
Monaco
US
United States of America
n
Finland
MG
Madagascar
FBI
France
ML
Mali
ISDOCID: <WO 8607174A1J_>
WO 86/07174
PCT/US85/01435
1
SUPER-COMPUTER SYSTEM ARCHITECTURES
Field of the Invention
This invention relates to data processing systems
and, more particularly, to computer system architectures
which are especially applicable to large scale, powerful
systems •
Background of the Invention
Some current and many contemplated applications for
large scale r powerful computer systems require tremendous
capabilities for computation which continue to push the
state of the art in the field for current applications
and which far exceed the state of the art for
contemplated applications* The performance of computer
systems has generally been evolutionary in that the
fundamental architecture has remained in a traditional
configuration (sometimes called "Von Neumann") involving
the sequential execution of instructions which
individually are rigidly defined. Even such techniques
as parallel processing typically involve arrays of
traditionally configured processors functioning under the
coordination of a master processor. Virtually all known
present and contemplated (insofar as they are disclosed
in the literature) system architectures can be analyzed
and identified as Von Neumann variations.
One of the present trends for computer system
architectures is toward very fast processors having
relatively limited command structures. Thus, it is not
unlikely that most future computers will be more
elementary than those in current use, but will be
ultrafast. One significant drawback for this possible
evolutionary path is that more and more of the
"responsibility" for system performance falls on the
softwar , and it is the experienc of the industry that
the performanc of many fine computer systems remains
software limited. That is, the ultimate p rformance of
the systems are limited by inefficiencies which are
FCT/US8S/© 1435
2
simply unavoidable in all but vcsry @h@rt programs written
in machine language*,
Many future applications, such as in artificial in-
telligence, advanced space technology" and the like,
Impose computational requirements which exceed the
ability of traditional Von Neumann systems, no matter how
closely the theoretical ©peed limits are approached g and
no matter how configured or arrayed <>
Thus, it will be appreciated by those skilled in the
art that the fundamental architectural approach for large
scale 9 powerful systems must be rethought if demanding
future applications are to be dealt witho It will
therefore be commensurately appreciated by those skilled
in the art that it would be highly desirable to provide a
new architectural structure by the use of which the
fundamental and inherent limits of conventional computer
systems may be avoided „
WO 86/07174
PCT/US85/0143S
3
Objects of the Invention
It is therefore a broad object of my invention to
provide a new computer system architecture.
In another and fundamental aspect, it is an object
of my invention to provide a computer system architecture
which admits of modification of the computer instruction
set.
It is a further object of my invention to provide a
computer system architecture in which the computer
instruction set can be modified by stored program
control .
It is a more specific object of my invention to
provide a computer system architecture in which each
instruction of the computer instruction set, whose
execution is coordinated by a program control computer,
is executed by an individual instruction computer.
It is a still further object of my invention to
provide means in each instruction computer for modifying
the effects of executing a given instruction prior to or
during execution.
.8607174A1_I_>
IPCT/US85/0143S
4
Summary of the Invention
These and other objects of my invention are achieved
by employing a unique computer system architecture in
which a program control computer is in communication with
a bank of instruction computers o The program control
computer includes instruction execution coordinating
mean© which respond to the presence ©f an instruction in
a program sequence by identifying the instruction and
selecting an instruction computer to execute the
instruction o Each instruction computer includes
programmable instruction decoding means which serve to
identify the microtasks required for executing a given
instruction o The programmable instruction decoding means
are further adapted to respond to instruction
modification signals applied thereto to change the
microtasks identified as necessary for executing the
instruction o The programmable instruction decoding means
have the ability to dynamically change the precise
execution of an instruction p not only between successive
execution cycles, but also during a single execution in
response to intermediate results 0 The power of the
system may be increased by extending levels in a two-
dimensional array and extending the multi-level systems
in a three-dimensional array 0
WO 86/07174
PCT/US85/0I43S
5
Description of tfrg pyavlnq
The subject matter of the invention is particularly
pointed out and distinctly claimed in the concluding
portion of the specification. The invention , however,
both as to organization and method of operation, may best
be understood by reference to the following description
taken in conjunction with the subjoined claims and the
accompanying drawing of which:
Fig. 1 is a ma j or block diagram of an exemplary
computer system employing the present architecture;
Fig. 2 is a more detailed block diagram of the
system of Fig. 1; and
Fig. 3 illustrates the extension of the system of
Figs. 1 and 2 into multi-level two-dimensional arrays and
three-dimensional arrays of multi-level systems.
,86071 74A1_L>
WO 86/07174
PCT/US85/ttl435
<S
Detailed Description of the Inventi n
The major block diagram of Fig 0 1 illustrates a
fundamental aspect of the present invention which sets it
apart fro® prior art data processing system
architectures o She system of Fig D 1 includes any
suitabi© aggregation of input/output devices and mediums
1 in communication with a high speed program control
computer 2o Also in communication with program control
computer 2 is a system main memory 3 which may consist of
any assemblage of memory storage means appropriate to a
given system application 0 Both the high speed program
control computer 2 and the system main memory 3 are in
communication with a bank of instruction computers 4„ As
few as one instruction computers could comprise the bank
4, but efficient embodiments of the invention preferably
include at least as many instruction computers in the
bank 4 as there are instructions in the repertoire of the
high speed program control computer 2 * Thus, the bank of
instruction computers 4 illustratively include
instruction computer (00) 5, instruction computer (01) 6,
instruction computer (02) 7, instruction computer (03) 8,
an indeterminate number of instruction computers 9 and
instruction computer (n) 10 which represents the final
instruction computer in the illustrative bank 4„
It will be understood that the high speed program
control computer 2 does not itself execute all the
instructions in a program executed by the system , but
rather assigns the execution of each instruction in its
instruction set to one instruction computer among the
bank 4 0 Merely by way of elementary example, if the
instruction °°00 m in the repertoire of the high speed
program control computer 2 is 00 add one to operand 00 , the
program control computer 2 may issue the operand to
instruction computer 00 5 along with a signal advising
the instruction computer 00 to A) initiate instruct! n
execution, B) advise the high speed program control
computer 2 when execution of the instruction has been
WO 86/07174
PCT/US85/01435
7
completed (or some m aningful intermediat result has
been btain d) , and C) r turn the r suit to the high
speed control computer 2. Other instructions in the
repertoire of the high speed program control computer 2
may be executed (serially, in parallel, or both) by the
instruction computers comprising the bank 4.
A second fundamental aspect of the architecture of
the system comprising the present invention is that, as
will be discussed more fully below, means are including
in each instruction computer in the bank 4 for altering
the meaning and consequent execution of an instruction to
be performed (or being performed) by an instruction
computer. Thus, returning to the elementary example
given above, the instruction M 00", which would ordinarily
be assigned for execution to instruction computer (00) 5,
might have evolved within the instruction computer "OO"
from an "add one to the operand" to something very much
more complex (still by way of example: a macro-
instruction such as "find the prime numbers falling
between limits a and c") which would better serve
performance of the program segment under immediate
execution, and the change might be effected within the
instruction computer (00) either before or during
instruction execution. The source for directing the
change in the execution of an instruction may originate
from the system program being performed, from human
intervention as through an operators 1 console or from
within the assigned instruction computer itself, as
through interpretation of intermediate results. Thus,
the instruction "00" might be further refined, between
executions or during a single execution from intermediate
results, to "find the second prime number between the
limits b and c, a < b < c". Thereafter, when the
instruction "00" is sensed during execution of a program
by the high speed program c ntrol computer 2, that
instruction will b so x cuted until again r vis d or
reinitialized.
86/4D7174
PCT/US8S/0M3S
8
Figo 2 is a more detailed block diagram of the
system illustrated in Fig«> 1 0 2n Fig<> 2 , the
input/output block has been separated into input la,
output lb, and operator® 0 console le 0 The program
control computer 2, the system main memory 3 and the
illustrated instruction computers 5,6,7,8, and 10 from
the bank are all coupled for mutual communication through
a comprehensive bus system 11 0 The operators 0 console 1c
also has access to the bus system 11, and it is
contemplated that some embodiments of the invention would
be most efficient if the input sub-system la and output
subsystem lb had direct access to the bus system 11 .
The input la and the output lb sub-systems are not of
direct import to the fundamental aspects of the present
invention and are therefore shown in Fig a 2 as
communicating only with the program control computer 2«
High speed program control computer 2 includes
instruction and operand register (s) 12 for receiving and
temporarily storing instructions and any associated
operands from, for example f the system main memory 3o An
arithmetic and logic unit 13 in the program control
computer 2 communicates with the instruction and operand
registers 12 , the bus system 11, and an instruction
execution coordinating unit 14 which performs supervisory
functions pertaining to execution instruction within the
system • The instruction execution coordinating unit 14
also accesses the bus system 11 and serves to continually
monitor and regulate the relationship between the program
control computer 2 and the instruction computers 0 The
instruction execution coordinating unit 14, upon
determining the identification of an instruction to be
executed in accordance with the currently running
program, determines the specific instruction computer to
which that instruction is currently assigned for
execution, determines the status (i.ee, availability or
potential availability) of the assigned instruction
computer and, when appropriate, causes an indication that
the instruction computer is to commence its execution
WO 86/07174
PCT/US85/01435
9
function operating n such operand information as may be
supplied t it via th bus system 11, as may be available
from the results of previous execution (s) or both.
The instruction computers comprising the instruction
computer bank are, for most contemplated applications,
preferably essentially identical to one another; however,
this is not a constraint on the inventive system arch-
itecture, and it is further contemplated that a hierarchy
of instruction computers of differing power may be more
appropriate for certain system applications.
Thus, instruction computer (00) 5 includes
instruction assignment and operand register (s) 15 for
receiving and temporarily storing the information
necessary to execute the instruction for which
instruction computer (00) 5 is directed and set up to
perform. On commencement of execution, a programmable
instruction decoding unit 16 is activated and issues
signals to instruction microtask generator unit 17. The
information contained in the signals applied to the
instruction microtask generator unit 17 from the
programmable instruction decoding unit 16 depends upon
the current decoding configuration of the latter. The
instruction microtask generator unit 17 responds to the
applied signals by issuing signals representing the
microtasks which must be performed in the system to
execute the instruction assigned to the instruction
computer (00) 5 as currently interpreted by programmable
instruction decoding unit 16 in instruction computer
(00). These microtask signals are applied, as may be
appropriate, to an arithmetic and logic unit 18 in the
instruction computer 5, an internal memory 19 in the
instruction computer, a status unit 20 in the instruction
computer and back to the programmable instruction
decoding unit 16 in the instruction computer. Such
microtasks, if any, as may be necessary for performance
outside the instruct! n computer 5 to complet
instruct! n ex cution are issued to the bus system 11 and
communicated to th ir d stination. Preferably, each
wo m/mi74
PCT/US8S/0M35
10
instruction computer in the bank is capable of generating
a comprehensive set of microtask signals to effect any
data manipulation of which the system is capable o Thus,
each instruction computer has the inherent ability to
perform powerful macro-instructions which may evolve
therein „
Arithmetic and logic unit 18, internal memory 19,
and programmable instruction decoding unit 16 may all
communicate among one another „ The arithmetic and logic
unit 18 also receives operand information from the
instruction assignment and operand register (s) 15, issues
signals to the status unit 20 and also is capable of
placing information on the bus system 11 for destinations
external to the instruction computer 5o
A primary feature of the programmable instruction
decoding unit 16 is its programmabilityo That is, it may
be reconfigured to issue a different set of signals to
the instruction microtask generator unit 17 whereby a
given instruction may be interpreted and executed
differently during different instruction cycles o Thus, a
programmable instruction decoding unit 16 may be
reconfigured under the influence, separately or in
conjunction with one another, of external signals from
the program control computer, system memory, or from the
console under the influence of a programmer «> From
internal sources, the programmable instruction decoding
unit 16 may be reconfigured by signals from the
arithmetic and logic unit 18, internal memory 19, and the
instruction microtask generator 17 o
It will therefore be understood that the actual
interpretation and execution of an instruction received
by an instruction computer may be varied from execution
cycle to execution cycle and, further, that its
configuration can be modified during a single execution
cycle as a result of the interpretation of intermediate
results or for other reasons which render the
00 adaptation ra of the precise execution of the instruction
to be desirable o
4SDOCID: <WO 8607174A1_I_>
WO 86/07174
PCT/US85/01435
11
The slgnificanc of the system architecture
presented h rein will now b come more apparent to those
skilled in the art . By this system architecture , not
only can the system, as a whole, "learn 19 to execute a
program more efficiently, but the instructions themselves
can be adapted between successive executions and even
during execution. Because of this ability to learn and
adapt, the spread of system responsibility between
software and hardware can be optimized; i.e., the system
becomes less software-bound since the software need not
be as complex and detailed as with traditional high speed
computer systems in which the "definition 19 of individual
instructions remains fixed or only slightly modifiable
within strict and predetermined limits.
The actual logical design of several constituents of
the system illustrated in Fig. 1 and 2, particularly
those of the program control computer 2 and the
instruction computers in the bank 4 as exemplified by the
instruction computer 5, straightforwardly follow their
function and can be carried out according to conventional
techniques. It may be noted that the operating systems
employed, respectively, in the program control computer
2 and in the instruction computers need not necessarily
be the same. An instruction computer need only be
furnished with operand-like information if necessary,
instruction modification information if applicable, and
an indication that it is to commence undertaking its
internally defined and adaptable instruction sequence and
have the results available when completed (or at an
intermediate point) as indicated by the status unit 20.
Therefore, the instruction computers, operating at a
different level from the program control computer, are
somewhat independent and can use an operating system
optimum for their structure as chosen by the logic and
circuit designers*
Communication (over the bus system 11 and otherwise
such as in dedicated channels) between the various system
constituents may b perf rmed in parallel, in seri s or
INSDOCIO: <WO 6607174A1_I_>
WO 86/07174
PCT/US85/0JL435
12
in a combination series/parallel aarmer as^ again, the
detailed design of an individual system may prescribe «
Th<& system shown in Fig c 2 is essentially uni-
dimensional in that there is shown only a single program
control computer, a single bank of instruction computers,
and & single level ©f main memory and input/output <>
However^, the system architecture is especially well
adapted for integration into very large scale two-
dimensional and three-dimensional supersystems
particularly including those subject to reconfiguration
under program control <> Referring now to Fig 0 3,
representations of such supersystems employing extensions
of the architecture of the present invention are
presented o Xt will be seen in Fig 0 3 that multi-level
system (0) 25 comprises a series of program control
computers 26 disposed in levels 0°p„ Similarly, a series
of instruction computer banks 27 are disposed at levels
0- m and a series of system memories 28 are disposed at
levels 0-1 o (Thus, each level of the multi-level system
25 comprises much of the structure illustrated in Fig„
2 o ) Communication both within a level and among
different system levels may be carried out across a
three-dimensional multi-bus system 29 » Input/output 30
may be coupled into the system at one or more levels,
typically interfacing either with the multi-bus system 29
or one ©r more program control computers o
The two-dimensional supersystem comprising multi-
level system 0 may be further extended by interfacing the
multi~bus system 29 with additional multi-level systems
1- g 31 o It will be apparent that, under program control,
the instruction computers at different levels in multi-
level system (0) 25 can be accessed by the program
control computers 26 and memories 28 at diverse levels in
order to achieve not only redundancy, but the ability to
adapt to a complex problem virtually to the extent of
realising the equivalent of a hardwired, completely
special purpose system., The provision of additional
SDOC1D: <WO 8607174A1J_>
WO 86/07174
13
PCT/US85/01435
multi-level syst ms 31, within practical limits, further
extends the system power to adapt.
in a multi-level system, one program control
computer must generally be dominant in order to direct
system reconfiguration as may be useful and to resolve
conflicts which may arise from, for example, attempts by
multiple program control computers to access a given
instruction computer. Similarly, in a three-dimensional
system including a plurality of multi-level systems, a
hierarchy must be established, and a single program
control computer will be the ultimate arbiter.
Referring again to Fig. 2, it will be noted that
system initialization is a substantial event because each
of the instruction computers must be made aware of its
initial instruction decoding configuration .
initialization may be performed by running, in the
program control computer 2, an initialization program
which assigns to each instruction computer in the
instruction computer bank its beginning instruction
sequence. Alternatively, an initialization signal may be
applied to each instruction computer which has
permanently stored therein (as in internal memory 19) the
key to its initial instruction decoding. For achieving
the most flexibility and for providing redundant
capability against the failure of one or more instruction
computers, the former procedure is preferred.
As previously noted, the logical design of
individual computer systems employing the novel
architecture set forth herein is susceptible to
performance using standard techniques and will vary
according to the system size, intended application, logic
family chosen, etc.
Therefore, while the principles of the invention
have now been made clear in an illustrative embodiment,
th re will be obvious, to those skilled in the art, many
modifications of structure, arrangements, proportions,
and the elements us d in the practice of th invention
which are particularly adapted for specific environments
NSDOCID: <WO B607174A1J_>
WO 86/07174
PCT/US85/01435
14
and op rating requirements without departing from those
principles.
SDOCID: <WO__ee07174A1_l_>
WO 86/07174
PCT/US85/01435
15
I CLAIM:
1. A data processing system including:
A. a program control computer;
B. a bank of instruction computers;
C. means coupling said program control computer and
each instruction computer in said bank for
communication therebetween;
D. instruction execution coordinating means
included in said program control computer and
adapted to transfer a signal representing an
identified instruction to one of said instruction
computers ; and
E. programmable instruction decoding means included
in at least one of said instruction computers for
receiving a signal representing an instruction and
for responding thereto by generating microtask
signals representative of the tasks required for
executing the instruction, said programmable
instruction decoding means being further adapted to
respond to instruction modification signals applied
thereto by changing the microtask signals generated
thereby*
2. The data processing system of Claim 1 in which each
instruction in an instruction set of said program control
computer is assigned to an instruction computer in said
bank and in which each said instruction computer in said
bank includes programmable instruction decoding means for
receiving a signal representing said assigned unique
instruction and for responding thereto by generating
microtask signals representing the tasks required for
executing said assigned unique instruction, each said
programmable instruction decoding means being further
adapted to respond to instruction modification signals
applied thereto by changing the microtask signals
gen rated thereby.
WO
®
PCT/US8S/0143S
16
3o Th© data processing system of Claim 2 in which said
program control computer includes means responsive to
system initialisation for generating a group of
predetermined instruction modification signal sets and
for transferring a predetermined one of said instruction
modification signal sets to said programmable instruction
decoding means in each said instruction computer in said
bank to establish initial mierotask signals generated
thereby in response to a signal representing said unique
instruction assigned to each said instruction computer 0
4o The data processing system of Claim 2 in which each
said instruction computer includes feedback means for
selectively responding to intermediate results obtained
during the execution of an instruction for issuing
instruction modification signals to said programmable
instruction decoding means included therein whereby
mierotask signals generated by said programmable
instruction decoding means are changed during instruction
execution o
5o The data processing system of Claim 1 which further
includes a main system memory ? in which said coupling
means comprises a bus system and in which said program
control computer? said main system memory? and each of
said instruction computers have access to said bus system
for communication thereamongo
So The data processing system of Claim 2 which further
includes a main system memory, in which said coupling
means comprises a bus system and in which said program
control computer? said main system memory? and each of
said instruction computers have access to said bus system
for communication thereamongo
7o Th data processing system of Claim 3 which further
includes a main system memory? in which said coupling
means comprises a bus system and in which said program
JSDOCID; <WO 86071 74A1_I_>
PCT/US85/01435
17
control comput r, said main system memory, and ach of
said instruction computers hav access t said bus system
for. communication thereamong*
8. The data processing system of Claim 4 which further
includes a main system memory, in which said coupling
means comprises a bus system and in which said program
control computer, said main system memory, and each of
said instruction computers have access to said bus system
for communication thereamong.
9. A data processing system including:
A. a plurality of program control computers;
B. a supervisory computer;
C. a plurality of instruction computer banks, each
of said instruction computer banks comprising a
plurality of instruction computers;
D. a plurality of system memories;
E. a bus system coupling said program control com-
puters, said supervisory computer, said system
memories and said instruction computers for
communication thereamong;
F. instruction execution coordinating means
included in each of said program control computers
and adapted to transfer a signal representing an
identified instruction to one of said instruction
computers ; and
6* programmable instruction decoding means included
in each of said instruction computers for receiving
a signal representing an instruction to be executed
and for responding thereto by generating microtask
signals representative of the tasks required for
executing the instruction, said programmable
instruction decoding means being further adapted to
resp nd to instruction modification signals applied
theret by changing the microtask signals generated
th reby.
WO 86/07174
<ISDOClD: <WO 8607174A1_I_>
PCT/US85/0M35
18
10 o 'Efoe data processing system of Claim 9 in ^hich said
supervisory computer is a selected one of said program
control computer^
11 o The data processing system of Claim 9 in ^hich each
said instruction computer includes feedback means for
selectively responding to intermediate results obtained
during the execution of an instruction for issuing
instruction modification signals to said programmable
instruction decoding means included therein whereby
microtask signals generated by said programmable
instruction decoding means are changed during instruction
execution <>
12 o The data processing system of Claim 10 in which each
said instruction computer includes feedback means for
selectively responding to intermediate results obtained
during the execution of an instruction for issuing
instruction modification signals to said programmable
instruction decoding means included therein whereby
microtask signals generated by said programmable
instruction decoding means are changed during instruction
execution*
WO 86/07174
PCT/US85/01435
1/2
SYSTEM MAIN MEMORY
INPUT/OUTPUT
DEVICES/ MEDIUMS
HIGH SPEED
PROGRAM CONTROL
COMPUTER
r2
1 1 t
INST.
COMP
00
INST.
COMP
01
INST.
COMP
02
INST.
COMP
03
BANK OF _
INST.
COMP.
It
INSTRUCTION COMPUTERS *
7
8
4
25
MULTILEVEL
SYSTEM O
—1 1
!\m
to
CM
o
M
-J
-J
_J
UJ i
flUJ
UI
UI
UI
UI
UI
f >
1 w
>
UI
>
UI
>
-J 1
A -J
-1
-J
>-
>-
DC
o
o
o
O
O
o
2
\s
2
2
2
ui
|UJ
UI
UI
UI
UI
UI
2
Is
2
2
2
2
2
[ SYSTEM
° J
SYSTEM
6 "
SYSTEM
5
SYSTEM
4
SYSTEM
3
SYSTEM
2
SYSTEM
1
INST. COMR-LEVEL 0
INST. COMR-LEVEL 1
?
INST. COMR-LEVEL 2
27
D
INST. COMR-LEVEL 3
INST. COMR-LEVEL 4
\
INST. COMR-LEVEL 5_
27
"HNSX COMP -LEVEL m
)
INPUT/OUTPUT
7
30
SUBSTITUTE SHEET
INSOOCID: <WO S607174A1_I_>
WO 86/07174
PCT/US85/0143S
2/2
SYSTEM
MAIN
MEMORY
IS
INPUT
OUTPUT
IB I*
OPERATOR'S
CONSOLE
A
PROGRAM CONTROL COMPUTER
'IS
INSTRUCTIONS OPERAND REGISTER IS J
flf
INSTRUCTION EXECUTION
COORDINATING UNIT
ARITHMETIC §>
LOGIC UNIT
£/L 2
BUS SYSTEM
£21
INSTRUCTION
COMPUTER 00
U INSTRUCTION !>
I COMPUTER 01 [
INSTRUCTION
COMPUTER 02
P~IO
INSTRUCTION
COMPUTER 03
INST.
COMRn
IS
z.
INSTRUCTION ASSIGNMENT AND OPERAND REGISTER IS)
FROM
FROM S YSTEM
MEMORY
FROM CONSOLE-
{18
PROGRAMMABLE
INSTRUCTION DECODING
SL
-IF
INSTRUCTION
MICROTASK
GENERATOR
ONTERWAL
MEMORY
IS
u
AROTHMETIC AND
LOGIC UNIT
rSO
STATUS
4SDOCID: <WO 8607 174A1_I_>
su® statute sheet
INTERNATIONAL SEARCH REPORT
International Application No PCT/US 85/01435
1. CLASSIFICATI
N OF SUBJECT MATTER (it several classification symbols apply, indicate all) *
According to International Patent C tees trie at ion (IPC) or to both National Classification and IPC
4 ^ G 06 F 9/38; G 06 F 15/06
II. FIILDS SEARCHED
Minimum Oocumantatfon Searched 7
Classification System
Classification Symbols
IPC 4
G 06 F
Oocumantation Saarehad othar than Minimum Documentation
to tht Extent that auch Documents are Included In the Fields Searched •
III. DOCUMENTS CONSIDERED TO RE RELEVANT*
Category *
Rstevant to Claim No. *>
Y
EP,
A, 0121763 (M.S. GREGORY et al.) 17
October 1984
see page 8, line 22 - page 12, line
14 and page 22, lines 13-34
1 ,2,4-6,8
Y
US,
A, 3308436 (W.C. BORCK et al . ) 7
March 1967
see column 3, line 3 - column 5, line
54
1 ,2,4-6,8
A
A
DE,
EP,
A, 2425380 (G. URSCHLER) 27 February
1975
see page 5, last paragraph; page 6;
page 12, paragraph 2 - page 20 ,
paragraph 1
A1, 0077404 (M. KURAKAKE) 27 April
1983
see page 5, line 13 - page 6, line 11
1 ,5
1-3
A
Proceedings of the 1983 International
./.
♦ Spatial categories of cited documents:
M A** document defining the general state ol the art which is not
considered to be of particular relevance
W E N earlier document out published on or attar the International
filing data
m L m document which may throw doubts on priority claim(s) or
which la cited to establish the publication date of another
Citation or other special reason (as specified)
"O" document referring to an oral disc to sure, use, exhibition or
ether means
*P N document published prior to the Intemstional filing date but
later than the priority date claimed
M T* later document published after the International flHng date
or priority date and not in conflict with the application but
cited to understand the principle or theory underlying the
Invention
"X" document of psrttcular relevance; the claimed Invention
cannot be considered novel or cannot be considered to
Involve an Inventive step
**Y" document of particular relevanco;* the claimed Invention
cannot be considered to Involve an Inventive etep when the
document le combined with one or mora other auch docu*
mente. such combination being; obvious to a person skilled
In the art.
"A* document member ol the same patent family
IV. CERTIFICATION
Date of the Actual Completion of the International Search
9th October 1986
Oate of Mailing of this International Search Report
2 f NOV 198ft
International Searching Authority
EUROPEAN PATENT OFFICE
Signature of Authoriied Off) eel
M van Mnr . _ £
Form PCT/ISA/210 (eecbnd sheet) (January IMS)
WSDOCID: <WO 8607174A1_I_>
Bntornotional Application Wo. p^T /US 85/014 3 "5 — 2 —
Coto0OC7 c
Citation of Oocumont, with mdicatcon, rcftoro opproorioto. of ttco rctovant pDsoogos
| Rolovant to Claim No
Conference on Parallel Processing
23-26 August 1983,, IEEE (New York,
Go Fritsch et aLs W EMSY 85 The
erlangen multiprocessor system for
broad spectrum of applications" ff
pages 325-330,, see page 326, left-
hand column
US)
9-12
Form PCT ISA.»210 (ostra ohool) (January 1CG5)
SDOCID: <WO B607174A1_L>
ANNEX TO THE INTERNATIONAL SEARCH REPOR^ON
INTERNATIONAL APPLICATION NO.
PCT/US 85/01435 (SA 13698)
This Annex lists the patent family members relating to the
patent documents cited in the above-mentioned international
search report «, The members are as contained in the European
Patent Office EDP file on 22/10/86
The European Patent Office is in no way liable for these
particulars which are merely given for the purpose of
information.,
Patent document
cited in search
report
Publication
date
Patent family
member ( s)
Publication
date
EP-A- 0121763
17/10/84
AU-A-
JP=A=
US-A-
CA-A-
US-A-
2527584
59223874
4580215
1209711
4546428
13/09/84
15/12/84
01/04/86
12/08/86
08/10/85
US-A- 3308436
GB-A-
DE-A-
FR-A-
1026890
1238695
1420405
DE-A- 2425380
27/02/75
GB-A-
AT-A, B
JP-A-
1456941
335202
50040243
01/12/76
25/02/77
12/04/75
EP-A- 0077404
27/04/83
WO-A-
JP-A-
8203708
57176456
28/10/82
29/10/82
For more details about this annex :
see Official Journal of the European Patent Office, No. 12/82
JSDOCID: <WO 8607174A1_I_>
THIS PAGE BUNK (usto)
WOSWW174
1/2
PCT/US8S/01435
INPLJT/OUTPUT
Devices/ 1 MEDIUMS
I
SYSTEM MAIN MEMORY
HIGH SPEED
PROGRAM CONTROL
COMPUTER
INST.
COMR
00
INST.
com
01
INST.
COMP.
02
INST
COMP
03
BANK OF
INST.
COMR
n
INSTRUCTION COMPUTERS m
25
MULTILEVEL
SYSTEM Q
[ SYSTEM
System
H
SYSTEM
5
SYSTEM
4
SYSTEM
3
SYSTEM
2
SYSTEM
t
31
MULT1- BUS
SYSTEM
INST. COMP-LEVEL 0
INST COMR- LEV EL t
INST. COMR-LEVEL 2
INST COMP-LEVEL 3
INST, OOMp- LEVEL 4
INST. COMP-LEVEL 5
INST OOMP
<4
-J
I
O
4
n:
a: la
inpitvoutput
_
40
B607174A1TL>
SUBSTITUTE SHEET
WO 06/01? 174
PCT7 USe3/m> 1435
2/2
DMOM
(INPUT
r
OPERATOR'S
OOMSOLE
IT
rnoiw comt^ol coqiifuTB^ ^
ua>®ie unit
£
ins syinsia
^ if
NSTRUCTUM
COMPUTER 03
Say
OWSTRajCTIliDK ASSIGNMENT AMD OPERAND RElEOSTgRl tSl
FROM
FROM g ?STgM
PROGRAMMABLE
MieHQfTASK
GENERATOR
MfeKNAL
m
u
aROTMftfllTlie AND
4
4
NSDOCID: <WO 8607174A1TI_>